Google Is Turning AI Search Into a Local Lead Engine

Google's May 20, 2026 ads announcement was written for marketers, retailers, universities, and ecommerce teams. Home-services operators should still read it closely.

Google is testing Gemini-built ad formats in AI Mode, expanding AI-powered Search ads, and introducing Business Agent for Leads. The details are still changing, but the direction is plain: Search is becoming a place where a customer can ask a complex question, compare options, get an AI explanation, click a sponsored recommendation, and start a lead conversation without moving through the old sequence of keyword, blue link, landing page, form.

For HVAC, plumbing, electrical, roofing, pest control, restoration, garage doors, and franchise service brands, this is not an ads-only update. It is a local proof problem.

The practical question is not whether Google will put more AI in ads. It already is. The practical question is whether each branch has enough proof for Google, customers, and AI systems to explain why that branch is the right provider for the job.

What Google changed on May 20, 2026

Google said it is testing two AI Mode ad formats built with Gemini: Conversational Discovery ads and Highlighted Answers. Conversational Discovery ads are designed to answer a person's specific question. Highlighted Answers make sponsored recommendations eligible to appear inside AI Mode recommendation lists. Google also said these formats will include an independent AI explainer and remain labeled as Sponsored.

The same announcement introduced Business Agent for Leads, a Gemini-built chat experience inside an ad. Google's example is a student researching universities, but the mechanism matters for home services: a buyer can click "Chat," ask questions, and get answers based on the advertiser's website before becoming a lead.

Google's Marketing Live collection framed AI Max as a way for businesses to become part of AI Search conversations. Google Ads Help says AI Max for Search campaigns can use broad match, asset-based matching, landing page-based technology, text customization, final URL expansion, locations of interest, and reporting that shows why ads matched.

That creates a new dependency. The ad system is no longer matching only against a keyword and serving a static page. It may choose a query-relevant URL, customize ad copy from landing page copy and assets, and use geographic intent controls.

For local-service brands, that means the landing page set matters more, not less.

Why home services has a different problem than ecommerce

Google's examples lean toward products, travel, shopping, and education. Home services behave differently because the customer is usually buying a local outcome under time pressure.

Someone searching for "best AC repair near me" is comparing risk as much as brand. They need to know who serves their neighborhood, who is open, who can send a technician, who has recent reviews, who handles the exact system, who has a working phone path, and who looks credible enough to let into the house.

Those facts rarely live in one place. Some live on the location page. Some live in the Google Business Profile. Some live in reviews. Some live in Yelp, BBB, Angi, HomeAdvisor, Facebook, YouTube, local news, or trade directories. Some live in technician performance that never becomes public evidence unless the business has a review operation.

That is why Google's AI Search guidance for local businesses matters alongside the ads news. Google says generative AI features in Search rely on core Search ranking and quality systems. For home services, the Search fundamentals are local fundamentals: accurate business data, crawlable service pages, complete profiles, reviews, photos, internal links, and third-party evidence that can be found and cited.

An AI ad can route demand. It still needs facts to route demand correctly.

The ad is only as good as the proof layer

AI Max and Business Agent for Leads make weak local data more expensive. If the campaign is allowed to expand into more queries, choose more URLs, customize more copy, and answer more questions, the system can expose every gap in the location structure.

A multi-location plumbing brand might have strong corporate copy but thin branch pages. A restoration rollup might have uneven Google Business Profiles after acquisitions. An HVAC platform might have thousands of five-star service moments, but the reviews are concentrated in three branches while newer markets look quiet. A garage door franchise might have accurate ads and messy citations.

The old campaign could hide some of that behind a landing page. The AI-assisted version is more likely to pull the gap into the conversation.

Before increasing AI Search spend, inspect these six assets for each priority market:

• The location page has the correct name, phone number, service area, hours, services, booking path, and local proof
• The Google Business Profile matches the page and uses complete categories, hours, services, photos, and review response practices
• The service pages answer real buying questions about urgency, process, cost drivers, equipment, guarantees, and eligibility
• The cited third-party sources show the same business identity and enough reputation evidence to support comparison queries
• The review operation asks customers neutrally and consistently at the point of service
• The tracking setup can survive dynamic landing pages, query-relevant URLs, and location-level attribution

This is where the paid and organic teams need the same operating model. The campaign needs good assets. The AI answer needs good sources. The branch needs good reviews. The operator needs to see which part is missing.

What Cheers is seeing in cited sources

Cheers' anonymized home-services checks from May 18 to May 20, 2026 point to one practical issue: AI answers are not building local recommendations from brand websites alone. The sample covered 13 home-services organizations, 113 tracked prompts, 13,797 provider results, and 120,928 valid source-domain mentions across ChatGPT, Perplexity, Gemini, and GAIO.

After excluding monitored business domains where matching was possible and applying a 10-organization threshold, the recurring domains were Yelp, Reddit, Google, Angi, Today's Homeowner, BBB, YouTube, HomeAdvisor, Facebook, and BestProsInTown.

That list should not be treated as one generic directory checklist. Yelp, Google, BBB, Facebook, and BestProsInTown usually carry reputation and location proof. Reddit adds informal customer language, comparisons, and objections. Angi and HomeAdvisor show category presence inside lead marketplaces. Today's Homeowner and YouTube can add educational or media context around the service category.

The operator takeaway is simple: each source has a job. Reviews prove customer experience. Citations prove the entity. Location and service pages explain fit. Third-party sources corroborate the claim. A branch can look strong in Google and still be underrepresented in the sources AI systems use for comparison prompts. A company can have strong reviews and still create confusion if acquisitions, old names, and duplicate profiles split the entity graph.

For the broader source strategy, compare this with our breakdown of how different AI search engines cite different sources. The point is to manage cited sources, review velocity, Google Business Profile hygiene, and location-page quality from one operating scorecard. If those workstreams sit in separate vendor lanes, the brand can fix individual listings while the source graph stays inconsistent.

How home-services brands should respond

Do not respond by making dozens of thin pages for every AI query variation. Google specifically warns against SEO work that exists only for machines. The better response is to make each priority location and service more complete, more specific, and easier to verify.

For a 40-location HVAC brand, that means the Phoenix AC repair page should not read like the Dallas AC repair page with the city swapped. It should explain the service area, common system issues, scheduling options, emergency coverage, technician credentials, review themes, and what happens when a customer books. It should link naturally to the Phoenix location page and relevant service pages. It should match the Google Business Profile. It should point to the same entity as the third-party profiles AI engines already cite.

For a plumbing franchise, the emergency plumbing page should answer the expensive questions: when to shut off water, what counts as an emergency, how pricing is handled, whether a branch offers after-hours service, what neighborhoods are covered, and how fast the customer can talk to a person. Those details are useful to customers first. They also give AI systems and ad agents something concrete to explain.

For a PE-backed rollup, the work starts with governance. Acquired brands, older domains, inconsistent names, duplicate listings, and uneven review volume create ambiguity. The citation stack and review collection process need owners, thresholds, and cadence. Sierra Cooling is the home-services example we can point to publicly: the case study shows how a stronger frontline review system and citation consistency work together for an HVAC business, without pretending the same result is automatic for every company.

What to measure before increasing spend

The wrong way to use the Google announcement is to ask only, "Should we turn on AI Max?" The better question is, "Which locations have enough evidence for AI Max, AI Mode, and Business Agent for Leads to represent them accurately?"

Start with market-level visibility. Pick the services and markets that matter most, then test how Google, ChatGPT, Gemini, Perplexity, and AI Overviews describe the options. Record which competitors appear, which sources are cited, which branch is chosen, and which facts are missing. A one-location snapshot from the Cheers AI Visibility Grader can show the shape of this problem, but a multi-location brand should turn it into a recurring market review.

Then map paid media readiness against the same evidence. If AI Max sends a buyer to the most relevant URL, is that URL actually the best page for the job? If Business Agent for Leads answers from the website, does the website contain the answer? If an AI Mode ad appears beside an independent explainer, do third-party sources support the claim?

Brands with specific local evidence will be easier to represent than brands with generic AI content. The advantage goes to locations that are easiest to understand, easiest to verify, and easiest to contact when the buyer is ready.

Methodology

Cheers used read-only aggregate Supabase queries over first-party AI visibility checks. The public sample used in this article covers May 18 through May 20, 2026 and includes home-services businesses only. It includes 13 home-services organizations, 113 tracked prompts, 13,797 provider results, and 120,928 valid source-domain mentions across ChatGPT, Perplexity, Gemini, and GAIO. We excluded monitored business domains where matching was possible and named only domains that appeared across at least 10 organizations.

The data does not include raw prompts, customer names, private business names, personal data, raw model outputs, or row-level results. It should be read as a current snapshot of source-domain appearances in Cheers checks, not as a universal ranking factor study.

Sources

• Google: A new generation of ads for the AI era of Search. Google's May 20, 2026 announcement for Gemini-built AI Mode ad formats, Business Agent for Leads, Direct Offers, and AI-powered Search ads
• Google Marketing Live 2026 collection. Google's overview of AI Search ads, AI Max, Ask Advisor, measurement, commerce, and creative announcements
• Google Search Central: optimizing for generative AI features on Search. Google's guidance that AI Overviews and AI Mode rely on Search fundamentals, useful content, technical access, and people-first quality
• Google Ads Help: how AI Max for Search campaigns works. Details on search term matching, text customization, final URL expansion, locations of interest, brand controls, and reporting
• Google: how AI Mode is changing the way people search in the U.S.. Google data on AI Mode query growth, longer questions, voice and image use, planning, and decision-oriented searches
• Google Search: a new era for AI Search. Google's Search I/O 2026 update on AI Mode, agentic booking, local services, and calls on behalf of users
• Google Business Profile Help: improve local ranking. Google's local ranking guidance on complete information, reviews, relevance, distance, and prominence
• Cheers: Google's AI Search guide for local businesses. Internal companion guide translating Google's AI Search guidance for local-service operators
• Cheers: AI search engines cite different sources. Internal source strategy guide for ChatGPT, Gemini, Perplexity, and AI visibility work
• Cheers: Sierra Cooling case study. Approved home-services proof point for review capture, citation consistency, and HVAC visibility work

Dylan Allen-Arnegård is the CEO & Co-Founder of Cheers, the local search platform for multi-location service businesses.